Ebolaviruses and marburgviruses (family Filoviridae) are negative-sense, single-stranded RNA viruses that cause severe hemorrhagic fever in humans and non-human primates. Ebola virus disease (EVD) and Marburg virus disease (MVD) are characterized by rapid person-to-person transmission, high case fatality rates, and a lack of approved treatments or vaccines. The family Filoviridae is divided into three genera [1]: Ebolavirus, Marburgvirus, and the genus Cuevavirus, recently discovered in a European bat [2]. Genus Marburgvirus contains a single species, Marburg marburgvirus, with two virus members, Marburg virus (MARV) and Ravn virus (RAVV), which are approximately 20% divergent at the nucleotide level [3]. The genus Ebolavirus includes five species, each of which contains a single virus member: Sudan ebolavirus (Sudan virus, SUDV), Zaire ebolavirus (Ebola virus, EBOV), Bundibugyo ebolavirus (Bundibugyo virus, BDBV), Taï Forest ebolavirus (Taï Forest virus, TAFV), and Reston ebolavirus (Reston virus, RESTV). Cuevavirus consists of a single species and virus, Lloviu cuevavirus (Lloviu virus). Historically, disease caused by MARV and EBOV have exhibited the highest fatality rates (up to 90% in some outbreaks), followed by SUDV (42%–65%), [4,5,6] and BDBV (36% to 40%) [7,8,9]. TAFV has caused one non-fatal human infection and RESTV is considered non-pathogenic to humans, but both can be highly pathogenic in nonhuman primates.

MVD was first identified in 1967 in Germany and the former Yugoslavia, when laboratory workers acquired a fatal illness after exposure to primates imported from Uganda [10]. EVD was first recorded in 1976 in Zaire (now Democratic Republic of the Congo, DRC) and South Sudan, during concurrent but unrelated outbreaks caused by EBOV and SUDV [11,12]. Since that time, sporadic outbreaks of both MVD and EVD have been recorded, usually involving dozens to hundreds of cases in remote locations in sub-Saharan Africa. The largest ever outbreak of SUDV, and until recently the largest outbreak of any filovirus, occurred in the Gulu district of Uganda in 2000–2001 and involved 425 cases and 224 deaths [13]. The current EBOV outbreak in Western Africa, which surpassed 25,000 cases in March of 2015 [14], represents a significant expansion of case numbers and a new geographic range for the virus, and clearly demonstrates the potential of filoviruses to become significant threats to public health on a global scale. TAFV was discovered in 1994 in Côte d’Ivoire, associated with mortality in wild chimpanzees and one non-fatal human infection [15,16]. RESTV has only been found in the Philippines, or in macaques imported from the Philippines [17,18,19]. Human exposures to RESTV have resulted in seroconversion without clinical signs of disease [20,21]. The most recently discovered ebolavirus, BDBV, was identified in 2007 in Western Uganda, and emerged again in 2012 in DRC [7,9].

The cave-roosting Egyptian rousette bat (Rousettus aegyptiacus, also called the Egyptian fruit bat), has been identified as a natural reservoir host for marburgviruses and consistent source of virus spillover to humans [22,23]. This discovery was based on identification of marburgvirus RNA and immunoglobulin G (IgG) [24,25] and the isolation of infectious marburgviruses [22,23,26] from wild rousettes inhabiting caves where human cases had recently occurred. Longitudinal studies have also demonstrated an association between the risk of human infection and the seasonal pulses of active marburgvirus infection in juvenile Egyptian rousettes during biannual reproductive cycles [23]. Cumulative evidence suggests various bat species also play a role in the transmission cycle of ebolaviruses. Epidemiologic links between ebolaviruses and fruit bats were identified in the first SUDV outbreak in 1976 [27] and in TAFV-associated disease in chimpanzees in Côte d’Ivoire in 1994 [15]. Detection of EBOV-specific IgG and, for the first time, RNA was reported in 2005 in fruit bats of three species (Hypsignathus monstrosus, Epomops franqueti, and Myonycteris torquata) hunted for food in Gabon and the Republic of the Congo (RC) [28]. Subsequently, an investigation into a large EBOV outbreak in DRC in 2007 showed a possible link between regional EVD re-emergence and seasonal fruit bat migration [29]. Since that time, several field studies have demonstrated sero-reactivity to EBOV antigen in a variety of fruit bat species, including the Egyptian rousette, in Ghana, Gabon, and RC [28,29,30,31]. Reactivity to recombinant RESTV nucleoprotein were reported in fruit bats in the Philippines [32] and Bangladesh [33], and in eleven different species of insectivorous and fruit bats in China [34]. However, in contrast to results for marburgviruses, repeated attempts at isolation of infectious ebolaviruses from bats have been unsuccessful.

Two recent experimental infection studies of Marburg virus in Egyptian rousettes have demonstrated virus replication in blood and multiple tissues [35,36]; oral shedding of infectious virus [36]; and viral antigen in liver and spleen without evidence of significant disease, findings which are consistent with expectations for a reservoir host. Though numerous field studies have demonstrated potential associations between bats and ebolaviruses, only a single experimental ebolavirus infection study has been conducted in any bat species [37]. In that experiment, a wide range of possible plant, invertebrate, and vertebrate hosts including insectivorous bats of two species (Mops condylurus, Chaerephon pumilus) and fruit bats of one species (Epomophorous wahlbergi) were inoculated with EBOV. Following inoculation, virus was successfully isolated from pooled viscera and blood from bats for up to three weeks, and was isolated from feces in one bat. There was also limited immunohistochemical staining for ebolavirus antigen in pulmonary endothelial cells in one insectivorous bat, without evidence of associated lesions [37]. Recently, a colony of Mops condylurus bats was found near the reported index case of the current Western African Ebola virus disease outbreak [38].

EBOV antibodies have been detected in wild Egyptian rousette bats in Gabon [29], and a R. aegyptiacus-derived cell line was shown to support EBOV replication in vitro [39]. Bats of other Rousettus spp. have been seropositive for RESTV and EBOV in the Philippines and China [32,33,34]. However, the capacity for Egyptian rousettes to become infected with ebolaviruses and act as a potential source of infectious virus is not known. Here, we report the findings of an experimental inoculation study of Egyptian rousette bats in which we compare the viral kinetics, tissue and cell tropism, potential for viral shedding, and clinical and pathologic effects of all five known ebolaviruses with findings from Marburg virus. This was a two-part study, consisting of a pilot study to investigate all six filoviruses concurrently, followed by a serial euthanasia study to compare the effects of SUDV infection with our previous findings for MARV. We hypothesized that, if Egyptian rousettes are not a true reservoir host of any of the five species of ebolavirus, then the response of this bat species to experimental infection with ebolaviruses will differ significantly from the response to Marburg virus infection. Inoculation of Egyptian rousette bats with ebolaviruses would result in either (1) abortive infection due to lack of susceptibility; or (2) clinical and pathologic signs of severe disease. We show that Egyptian rousettes are generally refractory to ebolavirus infection and are unlikely to act as sources of infectious virus in nature.

This is the first reported experimental infection study comparing the viral kinetics, tissue and cell tropism, and clinical and pathologic effects across six different filovirus species, in a bat host known to act as a natural reservoir for Marburg virus. The pilot study, in which four animals each were inoculated with identical doses of SUDV, EBOV, BDBV, TAFV, RESTV, and MARV, showed that tissue dissemination of ebolaviruses was limited in Egyptian rousettes, viremia was not detected, and there was no evidence of viral shedding via oral or fecal routes. In contrast, Marburg virus was detected in the blood, in a wide range of tissues, and in oral and rectal swabs of Egyptian rousettes in this study and in previous experiments [36]. These findings suggest that Egyptian rousettes are generally refractory to ebolavirus infection, implying they are not likely to act as a natural ebolavirus reservoir despite the identification of EBOV-seropositive Egyptian rousettes in Gabon [29] and RESTV-seropositive Asian bats of the Rousettus amplexicaudatus and R. leshenaulti species [32,33,34]. Furthermore, the Egyptian rousette, which tends to breed well in captivity and can thrive in a laboratory setting, may not be an appropriate experimental model for investigating ebolavirus-reservoir host relationships.

For Sudan virus, pilot study findings were intermediate: Viral RNA was more widespread than in the other four ebolaviruses, and was detected in both liver and spleen, though animals did not become viremic, viral loads were low, and SUDV antigen was very limited in distribution. These results were replicated and confirmed in a larger serial euthanasia study, which was designed to complement our previous Marburg virus serial euthanasia study [36]; day-by-day comparison of viral RNA levels in key tissues in SUDV and MARV serial euthanasia studies is provided in Figure 8.

SUDV RNA was detected in 10 different tissues, most frequently and at highest loads at 3 and 6 DPI. All 15 bats in the SUDV serial euthanasia study were PCR-positive in at least one tissue between 3 and 15 DPI, but viremia and viral shedding were not identified, and liver and spleen remained IHC-negative. In contrast with Marburg virus, which is frequently found in liver and spleen at levels consistent with replication in these sites, SUDV RNA was only detected in liver or spleen in 3 of 15 bats, and only one bat was PCR-positive in both tissues (Figure 8). The only tissues in which SUDV levels were suggestive of viral replication (TCID₅₀/g equivalents greater than inoculation dose) were the inoculation site at days 3 and 6, and the spleen in one bat on day 3. The presence of limited SUDV replication and relatively widespread tissue distribution (though at low levels) indicates that Egyptian rousettes could be more broadly susceptible to infection with Sudan virus than with other ebolaviruses, perhaps given a higher inoculum dose or different route of infection. However, the generally low tissue levels of viral RNA and the lack of any evidence of viral shedding suggest the virus would not be likely to persist in the population. No SUDV outbreak has ever been associated with caves or mines inhabited by Egyptian rousettes. Moreover, SUDV-specific RNA or antibodies have never been identified in any bat species, and the natural reservoir for SUDV remains undiscovered.

Bats inoculated with ebolaviruses did not display any clinical signs or hematologic changes consistent with significant disease, and histologic lesions were minimal. In the pilot ebolavirus study, two individual bats inoculated with either TAFV or RESTV had elevated total white blood cell counts, lymphocytes, and monocytes on two days each. Given that neither bat became viremic and no significant lesions were identified at necropsy, any relationship to viral infection was considered to be unlikely. However, since both bats’ WBC counts had returned to the normal range prior to euthanasia, it is possible that lesions were no longer present at necropsy. CBC values in the SUDV serial euthanasia study remained within normal limits for all bats.

AST was significantly elevated in one BDBV bat at 5 DPI, relative to controls and all other groups, with no associated histologic lesion, CBC abnormality, or significant weight loss. Similarly, in the SUDV serial euthanasia study, AST (but not ALT) was also significantly elevated at 3 DPI relative to controls and to any other day. Increased AST can be caused by liver damage, but also by damage to muscle or erythrocytes, and, in many species AST is less liver-specific than ALT [43]. In other megachiropteran bats, chemical and manual restraint have been shown to be associated with changes in blood chemistry values [44], and restraint-associated myopathy was speculated as a possible cause of increased AST in a survey of wild flying foxes (Pteropus giganteus). In this study, it is possible that AST elevation reflects restraint- or capture-associated myopathy rather than leakage from damaged hepatocytes. Creatine kinase (which specifically reflects muscle damage) was not measured, so was not available for correlation.

The Egyptian rousette is a natural host for marburgviruses, and a known source of virus spillover to humans. Unlike marburgviruses, no infectious ebolavirus has ever been isolated from a bat. Evidence supporting a role for bats as reservoir hosts for ebolaviruses is based primarily on ecological and epidemiologic data, which have demonstrated spatiotemporal association and epidemiologic links between human cases of EVD and bats. Though EBOV- and RESTV-seropositive bats have been found in areas where filoviruses have never yet been identified (for example, China [34]), ebolaviral RNA has been detected in bats in only a single study [28]. In our SUDV serial euthanasia study, we showed that two animals developed low SUDV titers without shedding virus, becoming viremic, or supporting widespread viral replication. Thus, though experimental inoculation was sufficient to induce seroconversion, there was no corroborating evidence to support this bat as a likely SUDV reservoir. Similarly, though field serosurveys have identified EBOV-seropositive Egyptian rousettes, these bats were generally refractory to EBOV infection in our pilot study. In contrast, experimental infections of Egyptian rousettes with MARV in this and previous studies [35,36] have replicated many features of natural MARV infections [22,23,25,26], and have expanded our understanding of the virus-reservoir host dynamics.

In conclusion, we have shown that Egyptian rousette bats are not likely to act as sources of ebolavirus spillover in nature. Indeed, the most likely bat candidates for ebolavirus reservoirs are the three species in which both ebolaviral IgG and RNA have been detected (Epomops franqueti, Hypsignathus monstrosus, and Myonicteris torquata) [28]. Our results, in particular the contrasts between ebolaviruses and Marburg virus, suggest the possibility of a one virus-one host species relationship, analogous to that in hantaviruses and rodent species.